T. McInerney and D. Terzopoulos. A finite element model for 3D shape reconstruction and nonrigid motion tracking. In Proc. Int. Conf. on Computer Vision (ICCV)'93, pages 518-523, 1993.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....triangulations by irregular ones that can have a few large facets in low curvature areas and more smaller facets in high curvature ones. It has been shown that this can be done within the deformable model framework by replacing the finite differences estimators by a finite element representation [7]. The computation of K would be different but the rest of the approach would remain. 4 The segmentation process We illustrate the process on the example of the putamen, a relatively small nucleus of the deep gray matter of the brain. The original binarized dataset is shown on Figure 5a) The ....

T. McInerney and D. Terzopoulos. A finite element model for 3D shape reconstruction and nonrigid motion tracking. In Proceedings of the Fourth International Conference on Computer Vision, Berlin, Germany, pages 518--523, May 1993.

.... by irregular ones that can have a few large facets in low curvature areas and more smaller facets in high curvature ones [20] It has been shown that this can be done within the deformable model framework by replac ing the finite differences estimators by a finite element representation [21, 22]. Using such a representation, the deformation energy ED of Equation (6) can still be computed as a quadratic form involving a stiffness matrix K. The computation of K would be different but the rest of the approach would remain. Generalizing this approach to handle fully irregular meshes should ....

T. McInerney and D. Terzopoulos. A finite element model for 3D shape reconstruction and nonrigid motion tracking. In Proceedings of the Fourth International Conference on Computer Vision, Berlin, Germany, pages 518--523, May 1993.

....in order to decrease the model sensitivity to small responses of the gradient operator. Nevertheless, the simplex mesh formalism allows the denition of regularizing constraints without any shrinking eoeect. RR n# 4078 20 Montagnat, Delingette 4.2. 3 Gradient force and boundary force Many authors [31, 53, 19] use both boundary and local gradient information. The external force is then a combination of these two terms: f ext (p) f gradient (p) f boundary (p) where f gradient is derived from the same potential eld as described above and f boundary is attracting vertices towards their closest ....

....scanning algorithm extracts a discrete voxel line along the model normal direction in a Cartesian image (left) or a cylindrical image (right) 4.4 Gradient force Gradient dioeerential operators are the most common method to extract boundary voxels [82] from volumetric images. A common method [31, 53, 19] uses a distance map of strong contours or edge points, extracted by thresholding the response of the gradient operator. In this paper, we do not rely on these edge points or additional gradient potential force f gradient . to dene boundary voxels. Indeed f gradient is negligible when a vertex is ....

T. McInerney and D. Terzopoulos. A Finite Element Model for 3D Shape Reconstruction and Nonrigid Motion Tracking. In International Conference on Computer Vision (ICCV'93), pages 518523, Berlin, Germany, May 1993.

....consider using the 3D array V shown in Figure 2 (b) as a real 3D data such as Magnetic Resonance Imaging (MRI) data or Computed Tomography (CT) data. This gives us the possibility of using a volume segmentation algorithm such as the one used by Cohen and Cohen [6] or by McInerney and Terzopoulos [16], which are based on deformable models. However, the assumption of considering the 3D array V as an MRI image is not valid because inherently V holds only the depth in Depth (b) Camera Side Z X Y (a) Figure 2: a) A rectified stereo pair (b) Visualization of the 3D array V filled by ....

T. McInerney and D. Terzopoulos. A finite element model for 3d shape reconstruction and nonrigid motion tracking. In ICCV93, pages 518--523, 1993.

....3D active shape models [48] These models utilize a predefined structure that incorporates prior knowledge about a shape s smoothness and its resistance to deformation. A number of different 3D deformable model formulations have been proposed; e.g. deformable tubes [32, 48] ellipsoidal models [9, 30], superquadrics [31, 36] etc. Perhaps the major limitation of such methods is the requirement that every object be described as the deformations of a single prototype object. This limits the kinds of shapes (and topologies) that can be recovered in general, since we can only recover shapes that ....

....that can be recovered in general, since we can only recover shapes that are achievable via the specific geometric model and nonrigid motion formulation. Some researchers attempt to overcome this limitation through the use of more general, 3D deformable part decompositions [35] local deformations [30, 31, 34], shape evolution models [13] or adaptive subdivision [21, 23, 52] These methods offer greater generality, but are still somewhat limited in the shapes and deformations they can describe in general. Furthermore, these techniques sometimes require careful initial placement of the model, reliable ....

T. McInerney and D. Terzopoulos. A finite element model for 3D shape reconstruction and nonrigid motion tracking. In ICCV, 1993.

....Both computer graphics and computer vision have to deal with surface modeling. The notion of a generalized cylinder was first introduced by Binford in computer vision long time ago and has been widely used in computer graphics. Terzopoulos s physics based models, such as finite element model, [11], adaptive meshes [17] and deformable superquadrics [16] can be directly applied to animation in computer graphics. The oriented particle surface model developed by Szeliski et al. 15, 14] is useful in both computer vision and computer graphics. We attempt to integrate computer vision with ....

T. McInerney and D. Terzopoulos. A finite element model for 3d shape reconstruction and nonrigid motion tracking. In Proc. 4th International Conference on Computer Vision, pages 518--523. IEEE, 1993.

....the predefined shape models provide extra constraints which incorporate prior knowledge about a shape s smoothness and its resistance to deformation. A number of different 3D deformable model formulations have been proposed and actually used for the task of structure and nonrigid motion recovery [19, 15, 13, 14, 8]. Terzopoulos et.al. 19] proposed a physically based modeling framework and recovered limited global descriptions (e.g. symmetry axis shape) from nonrigid motion. Pentland et.al. 15] used finite element method (FEM) models which incorporate elastic properties of real materials for the recovery ....

....local patch analysis. Global constraints are necessary in order to limit possible nonrigid behaviors and regularize the locally tracked nonrigid motion and structure. In the literature of computer vision, different kinds of shape models have been proposed as global descriptions of nonrigid motion [10, 19, 15, 13, 14, 8]. With global shape models, specific global constraints can be defined to guide the local nonrigid motion analysis by rejecting bad hypotheses (point with wrong structure or wrong motion) A well suited global shape model will be very beneficial for the problem of structure and nonrigid motion ....

T. McInerney and D. Terzopoulos. A finite element model for 3d shape reconstruction and nonrigid motion tracking. In ICCV93, pages 518--523, 1993.

....the model s generality but this leads to prohibitive computation time and numerical stability problems. In this paper, we consider the registration template as a surface model M to which a global transformation g is applied. The deformed model is thus g(M) fg(V i )g i . The deformable models [7] framework gives many more degrees of freedom to the model by applying a deformation field acting independently on each vertex. Data driven and regularizing forces are applied on the model. They result in a deformation field (a set of displacement vectors) L = fl i g 1in . We note the application ....

T. McInerney and D. Terzopoulos. A Finite Element Model for 3D Shape Reconstruction and Nonrigid Motion Tracking. In International Conference on Computer Vision, pages 518--523, 1993.

....3D active shape models [48] These models utilize a predefined structure that incorporates prior knowledge about a shape s smoothness and its resistance to deformation. A number of different 3D deformable model formulations have been proposed; e.g. deformable tubes [32, 48] ellipsoidal models [9, 30], superquadrics [31, 36] etc. Perhaps the major limitation of such methods is the requirement that every object be described as the deformations of a single prototype object. This limits the kinds of shapes (and topologies) that can be recovered in general, since we can only recover shapes that ....

....that can be recovered in general, since we can only recover shapes that are achievable via the specific geometric model and nonrigid motion formulation. Some researchers attempt to overcome this limitation through the use of more general, 3D deformable part decompositions [35] local deformations [30, 31, 34], shape evolution models [13] or adaptive subdivision [21, 23, 52] These methods offer greater generality, but are still somewhat limited in the shapes and deformations they can describe in general. Furthermore, these techniques sometimes require careful initial placement of the model, reliable ....

T. McInerney and D. Terzopoulos. A finite element model for 3D shape reconstruction and nonrigid motion tracking. In ICCV, 1993.

....Therefore, 4D data can be analyzed and compressed in a very efficient way. The power of the approach is illustrated by results on a 4D scan heart data. 1 Background For years, computer vision researchers have shown interest in nonrigid motion analysis using physicallybased deformable models [11, 5, 4]. Among these works, modal analysis, first introduced in computer vision by Pentland et al. 9] has the advantage of being a frequency based technique in which nonrigid motion is expressed in the free vibrations basis (modes) of the deformable object [7, 6, 8] More recently, many researchers ....

T. McInerney and D. Terzopoulos. A finite element model for 3-D shape reconstruction and nonrigid motion tracking. In IEEE Proceedings of the Fourth International Conference on Computer Vision, pages 518--523, Berlin, June 1993. IEEE.

....proposing improvements to a rigid body interpretation [26, 24] Following the theory of deformable models [10] physically based modelling for nonrigid motion analysis has become extremely popular. Terzopoulos et al. use a dynamic deformable model to infer the 3D structure of nonrigid objects [23, 11]. Duncan et. al use bending energy and surface curvature to analyze nonrigid motion [1] an alternative formulation can be found in [3] In [13] nonrigid motion is assumed to be locally conformal, while Chen and Huang use a tensor based analysis to estimate global motion parameters and local ....

T. McInerney and D. Terzopoulos. A finite element model for 3-D shape reconstruction and nonrigid motion tracking. In IEEE Proceedings of the Fourth International Conference on Computer Vision, pages 518--523, Berlin, June 1993. IEEE.

....techniques have been developed to capture the shape and motion of the inner or outer walls of an LV from medical image data. These models are constructed with finite element meshes derived from polyhedron surface reconstruction from a stack of cross sections [26, 36] physics based elastic models [27, 58, 47, 81, 67], bending and stretching models [1, 90] or augmenting local details to an axisymmetrical geometric model [11, 64] The techniques are briefly described below: ffl McInerney and Terzopoulos [58] developed the 3 D deformable balloon model [27] composed of triangular C 1 finite elements based on ....

.... reconstruction from a stack of cross sections [26, 36] physics based elastic models [27, 58, 47, 81, 67] bending and stretching models [1, 90] or augmenting local details to an axisymmetrical geometric model [11, 64] The techniques are briefly described below: ffl McInerney and Terzopoulos [58] developed the 3 D deformable balloon model [27] composed of triangular C 1 finite elements based on the framework developed by Metaxas and Terzopoulos [60] They applied it to CT data consisting of 16 volume images of a canine heart during one cardiac cycle. The reconstruction of the LV shape ....

[Article contains additional citation context not shown here]

T. McInerney and D. Terzopoulos. A finite element model for 3D shape reconstruction and nonrigid motion tracking. In Proc. 4th International Conference on Computer Vision (ICCV93), pages 518--523, Berlin, Germany, 1993.

....the nonrigid matching is quite related to the surface deformation methods because we deform the surfaces in order to bring the corresponding points nearer. For example, the deformation techniques presented in [BK89] Inria Rigid, Affine and Locally Affine Registration of Free Form Surfaces 5 [MT93], MT91] PS91] or [NA93] are very interesting. These authors deform the surfaces using a physical model involving internal and external forces. Our deformations are quite different: they are the result of a geometric transformation, and the constraints used are based on geometric differential ....

D. McInerney and T. Terzopoulos. A finite element model for 3d shape reconstruction and non rigid motion tracking. In Proceedings of the Fourth International Conference on Computer Vision (ICCV '93), Berlin, May 1993.

....the deformation as a variational problem involving an internal energy that enforces some continuity constraints, and an external energy controlling the closeness of fit. Several researchers have applied the dynamic model fitting scheme to range data or medical images[CCA92] DHI91b] VM93] Gue93][McI93][NA93] Lei93] Elastic models successfully address the problem of shape control. However, few researchers have proposed general adaptive reconstruction techniques for solving both geometric and topological aspects. In particular, surface models based on regular grids are not well suited for ....

....order to enforce C 1 or C 2 continuity across triangles (up to 18 nodal shape functions for C 1 continuity across triangles) Chen et al. CS93] fitted a progressively refined triangulation on range data and used GregoryB ezier patches for constructing a G 1 continuous representation. In [McI93], McInerney used a semi regular triangulated sphere for a finite element based deformable balloon, but provided only uniformed refinement capabilities. Vasilescu et al. VT92] developed an adaptive and discontinuity preserving algorithm for meshes of interconnected springs. Hoppe et al. HDDM93] ....

[Article contains additional citation context not shown here]

D. McInerney, T. Terzopoulos. A finite element model for 3d shape reconstruction and nonrigid motion tracking. In Proc. of the Fourth Int. Conf. on Computer Vision (ICCV'93), pages 518--523, 1993.

....motion from an image sequence. Most of these algorithms incorporate the elasticity assumption into a deformable model of the object, and fit the object model to the observed motion of the object s surface. Global motion parameters are then estimated from the object model motion. Terzopolous et al. [85, 86, 74, 87, 88] proposed a physics based approach, which uses deformable models of the object based on shape primitives. These models move in response to external force fields according to the Lagrange equations of motion. The force fields are used to fit the deformable model to the object in each frame of the ....

T. McInerney and D. Terzopoulos. A finite element model for 3d shape reconstruction and nonrigid motion tracking. In Proceedings of the Fourth International Conference on Computer Vision (ICCV'93), Berlin,Germany, 1993.

....section we define deformable intensity surfaces for image warping. 3. 1 Theoretical background Following the theory of active contour models [11, 26] several models have been developed that deal explicitly with deformable surfaces, among them : deformable superquadrics [19, 25] surface snakes [6, 13], particle systems [24] splines [3] and elastic thin plates [21, 18] The above models usually evolve in Euclidean 3D space, however, deformable templates which evolve in XY I space with application to feature extraction have been investigated by Yuille et al. [29] Hence, deformable intensity ....

T. McInerney and D. Terzopoulos. A finite element model for 3-D shape reconstruction and nonrigid motion tracking. In IEEE Proceedings of the Fourth International Conference on Computer Vision, pages 518--523, Berlin, June 1993. IEEE.

....to be taken into account, such as geometric constraints, noise, or uncertainty in the data. This basic problem arises in computer vision, robotic navigation, pattern recognition, data analysis, animation, computer simulation and elsewhere. There is a very large literature on this subject, e.g. [BW, Bo, Br, CG, EM, MT, HSIW]. The approaches to this problem can be characterized as either implicit or explicit. In the explicit approach one often uses a local spline approximation, where the knowledge of the ordering and connections between the data points is crucial. Alternatively one needs a global parameterization. ....

T. McInerney and D. Terzopoulos, A finite element model for 3D shape reconstruction and nonrigid motion tracking, Proc. IEEE 4th Int. Conf. on Comp. Vis., (1993), pp. 518-523.

....Both computer graphics and computer vision have to deal with surface modeling. The notion of a generalized cylinder was first introduced by Binford in computer vision long time ago and has been widely used in computer graphics. Terzopoulos s physics based models, such as finite element model, [4], adaptive meshes [10] and deformable superquadrics [9] can be directly applied to animation in computer graphics. The oriented particle surface model developed by Szeliski et al. 8, 7] is useful in both computer vision and computer graphics. We attempt to integrate computer vision with computer ....

T. McInerney and D. Terzopoulos. A finite element model for 3d shape reconstruction and nonrigid motion tracking. In Proc. 4th International Conference on Computer Vision, pages 518--523. IEEE, 1993.

No context found.

T. McInerney and D. Terzopoulos. A finite element model for 3D shape reconstruction and nonrigid motion tracking. In Proc. Int. Conf. on Computer Vision (ICCV)'93, pages 518-523, 1993.

No context found.

T. McInerney and D. Terzopoulos. A finite element model for 3d shape reconstruction and nonrigid motion tracking. In International Conference on Computer Vision, pages 518--523, 1993.

No context found.

Tim McInerney and Demetri Terzopoulos. A finite element model for 3D shape reconstruction and nonrigid motion tracking. In Proceedings of the International Conference on Computer Vision, pages 518--523, 1993.

No context found.

McInerney D., Terzopoulos T. A Finite Element Model for 3D Shape Reconstruction and Nonrigid Motion Tracking. In Proceedings of the Fourth Int. Conf. on Computer Vision (ICCV'93), pp. 518-523, 1993.

No context found.

Tim McInerney and Demetri Terzopoulos. A finite element model for 3D shape reconstruction and nonrigid motion tracking. In Proceedings of the International Conference on Computer Vision, pages 518--523, 1993.

No context found.

T. McInerney and D. Terzopoulos. A Finite Element Model for 3D Shape Reconstruction and Nonrigid Motion Tracking. In International Conference on Computer Vision, pages 518523, 1993.

No context found.

T. McInerney and D. Terzopoulos. A finite element model for 3-D shape reconstruction and nonrigid motion tracking. In Proc. ICCV, May 1993.

First 50 documents

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC